In linkPrediction example like ppa, the embeddings of nodes are firstly learned by node2vec and then used to predict links by MLP. Is the node2vec trained using all edges? If yes, this would contaminate the test data which is predicted by MLP (data leakage issue).
I have checked the edges of test data used by MLP model. It shows that the edges in test data are not included in the training data used by node2vec. Can you please confirm this?
In linkPrediction example like ppa, the embeddings of nodes are firstly learned by node2vec and then used to predict links by MLP. Is the node2vec trained using all edges? If yes, this would contaminate the test data which is predicted by MLP (data leakage issue).
I have checked the edges of test data used by MLP model. It shows that the edges in test data are not included in the training data used by node2vec. Can you please confirm this?